Intel to Protect Microchips from Cosmic Rays

As computers become more advanced, the microprocessors inside them shrink in size and use less electrical current. These new, energy efficient chips can be crammed closer together, increasing the number of calculations that can be done per second, therefore making the computer more powerful. But even the mighty supercomputer has its Achilles heel: an increased sensitivity to interference from charged particles originating beyond your office. These highly energetic particles come from space and may cause critical hardware to miscalculate, possibly putting lives at risk.

Foreseeing this problem, microchip manufacturer Intel has begun devising ways to detect when a shower of charged particles may hit their chips, so when they do, calculations can be re-run to iron out any errors…

Cosmic rays originate from our Sun, supernovae and other unknown cosmic sources. Typically, they are very energetic protons that zip through space close to the speed of light. They could be so powerful that on impact with the upper atmosphere of the Earth it has been postulated that they may create micro black holes. Naturally these energetic particles can cause some damage. In fact, they may be a huge barrier to travelling beyond the safety of Earth’s magnetic field (the magnetosphere deflects most cosmic radiation, even astronauts in Earth orbit are well shielded), the health of astronauts will be severely damaged during prolonged interplanetary flight.

But what about on Earth, where we are protected from the full force of cosmic rays? Although a small portion of our annual radiation dose comes from cosmic rays (roughly 13%), they can have extensive effects over large volumes of the atmosphere. As cosmic rays collide with atmospheric molecules, a cascade of light particles is produced. This is known as an “air shower”. The billions of particles within the air shower from a single impact are often highly charged themselves (but of lesser energy than the parent cosmic ray), but the physics behind the air shower is beginning to grow in importance, especially in the realms of computing.

It seems computer microprocessor manufacturer Intel has been pondering the same question. They have just released a patent detailing their plans should a cosmic ray penetrate the atmosphere and hit one of their delicate microchips. The problem will come when computing becomes so advanced that the tiny chips may “misfire” when a comic ray impact event occurs. Should the unlucky chip be hit by a cosmic ray, a spike of electrical current may be exerted across the circuitry, causing a miscalculation.

This may sound pretty benign; after all, what’s one miscalculation in billions? Intel’s senior scientist Eric Hannah explains:

All our logic is based on charge, so it gets interference. […] You could be going down the autobahn [German freeway] at 200 miles an hour and suddenly discover your anti-lock braking system doesn’t work because it had a cosmic ray event.” – Eric Hannah.

After all, computers are getting smaller and cheaper, they are being used everywhere including critical systems like the braking system described by Hannah above. As they are so small, many more chips can occupy computers, increasing the risk. Where a basic, one processor computer may only experience one cosmic ray event in several years (producing an unnoticed calculation error), supercomputers with tens of thousands of processors may suffer 10-20 cosmic ray events per week. What’s more, in the near future even humble personal laptops may have the computing power of today’s supercomputer; 10-20 calculation errors per week would be unworkable, there would be too high a risk of data loss, software corruption or hardware failure.

Orbital space stations, satellites and interplanetary spacecraft also come to mind. Space technology embraces advanced computing as you get far more processing power in a smaller package, reducing weight, size and cost. What happens when a calculation error occurs when a cosmic ray hits a satellite’s circuitry? A single miscalculation could spell the satellite’s fate. I’d dread to think what could happen to future manned missions to the Moon, Mars and beyond.

It is hoped that Intel’s plan may be the answer to this ominous problem. They want to manufacture a cosmic ray event tracker that would detect a cosmic ray impact, and then instruct the processor to recalculate the previous calculations from the point before the cosmic ray struck. This way the error can be purged from the system before it becomes a problem.

There will of course be many technical difficulties to overcome before a fast detector is developed; in fact Eric Hannah admits that it will be hard to say when such a device may become a practical reality. Regardless, the problem has been identified and scientists are working on a solution, at least it’s a start…

Source: BBC

15 Replies to “Intel to Protect Microchips from Cosmic Rays”

  1. The discussion about health issues in this article remind me of declarations in the 19th century that “everything that can be invented *has* been invented, so let’s close up the patent office and do other things.” Who knows what the future will bring, and what advances may be made that will allow Areonauts to travel to and live on Mars safely?

  2. Ok….so a trip to the moon is risky…and a trip to mars is a death sentence. Is that correct?
    If…and a very BIG if…this is true….it seems that space exploration truly is a future only for robots.

  3. Would the power requirements for generating a magnetic field around spacecraft be infeasible? if so how thick a sheet of water would suffice to protect astronauts from cosmic rays?

  4. It is worthwhile pointing out that the processors and other computer equipment that spacecraft use are already radiation hardened, so that these issues, although highly important in spaceflight, have been engineered against already:

    http://en.wikipedia.org/wiki/Radiation_hardening

    The issues talked about in this article are generally only valid for Earth-based computers that do not make design allowances for the fact that radiation strikes can occur and mess with computational equipment.

  5. One way to deal with this error problem is to use many computers rather than depend on one. As cosmic rays are rather small, a single ray striking atoms as it passes through only affects those, not the entire computer. If an error is generated it is limited to that chip or computer. With several computers doing the same work and cross-checking each others work, errors can be eliminated. In fact that’s how NASA is doing it now. Getting an error free computer is a case of chasing clouds anyway. Stray electrons (not comic ray generated) exist in computer chips caused from impurities in manufacturing materials used in making chips. Errors are generated. Computer programs are lead astray. Maybe thats why the government is wrong so often? Maybe the IRS is wrong and you really don’t owe that much in taxes. After all, your computer ran that tax program and it couldn’t be wrong – right?

  6. I think the main thing we need to remember that the identification of future problems – either the health risks, or computer faults – associated with cosmic rays is a good thing. Yes, some articles can be sensational (recent examples include scientists declairing that we should give up hope on Mars colonization, as radiation will be too high), but we’ve always risen to challenges and found answers.

    I remember a story when the first automobiles were engineered, there were serious concerns that going over 30 miles/hr would cause the driver to pass out due to lack of oxygen! They based this on the fact that turbulent eddies and pockets of vacuum will be created inside the cab. Yes, we have turbulent eddies, but they are far from dangerous, and besides, car designs are now highly aerodynamic, reducing the drag caused by eddies.

    So it goes to show that the initial fear of a potential problem is often sensationalised, only to find the problem isn’t that great a risk in reality. And besides, should there be a problem, we often find a solution very quickly. There is very little between *problem* and *solution* when us humans stand to gain from *progress*.

    Cheers, Ian

  7. You put it well, Yael.

    The comments and the article above also reminds of another rediculous scenario that Bob Zubrin likes to talk about- when airplanes first started taking passengers, people suffered the problem of altitude sickness. Naysayers started talking about how fast and high-flying airplanes will never be practical because humans cannot be comfortable in high-flying planes.

    A whole lobby of researches doing research on altitude sickness wanted their funding by the government for developing antidote for this problem kept harping about for decades about the relevance of their work for aeroplane companies and why the government should keep funding them- they will one day develop medicines which people flying planes could then take, ameliorating the low-pressure problem.

    Then one day, the private industry decided to just get rid of this rubbish- and decided to pressurize the cabins. Problem solved!

    The same is happeing today with the naysayers for manned space flight. It’s all about protecting the funding of people and groups who want to keep NASA flying in circles around Earth- and who want the money to keep coming in- for all sorts of frivolous reasons- researching effect of zero-gravity on humans and how to develop medicine for the same, effect of radiation on humans and computer chips and how to develop antidotes for those too.

    Soon, if companies such as Space-X succeed- they are literally going to suck the life out on such wastefulness and circus that has existed over the past 50 years and has hijacked humanity’s destiny beyond Earth. With a spinning spacecraft in space, and a spinning colony on Mars and Moon, they will get the gravity back, rather than support useless medicine-men, and put sheilding around spacecrafts and chips where needed.

  8. Why, according to the article and even your blog entry, Undistinguished, are some interstellar rays humourous? I haven’t ever laughed at a comic ray! Some are serious to be sure. But I think this is an issue we should be earnest about. Cosmic rays should be laughed at no more!

  9. I’ll reserve judgement until they actually DO something AND get results.

  10. How about writing more efficient software so we don’t have to put bleeding edge processors in things as mundane as antilock braking systems? It’s not like we needed dual-core processors to get to the moon. Oh, but that would decrease Intel’s sales… Sorry. Silly thought.

  11. SOme Intel old timers must be having a deja vu… in the early days of solid state memory technology (early 70’s) Intel had to cope with “Soft Errors” affecting random DRAM cells. These turned out to involve alpha particles from decaying radioactive atoms present in trace quantities in the chip packaging materials. Took them a while to discover what was causing it.

  12. “Ok….so a trip to the moon is risky…and a trip to mars is a death sentence. Is that correct?”

    No, it says what’s been known for some time: solid-state electronics aren’t that much better at tolerating ionizing radiation than living things are. The faster the components, the more sensitive.

    BOTH will require adequate shielding.

  13. I have no doubt that this problem will be solved (both for chips and for people).

    Note, however, that at least one well-known manufacturer of HD camcorders currently does not ship their product by air (surface/ship transportation only) specifically because of cosmic ray damage to the CCD and CMOS image chips while in transit. NASA has announced similar cosmic ray problems with their HD camcorders on the International Space Station. Source: IEEE International Symposium on Consumer Electronics 2007.

  14. Soft errors from alpha particles were well described in 1978 (May & Woods), from cosmic rays (JF Ziegler) in 1979, and thermal neutrons (R Baumann) in 1995. This has been a constant design issue for electronics in terrestrial and space applications. Each new generation of devices brings new challenges. The Intel news release is one more in a long series of engineering efforts.

  15. Funny that we are talking about this now for intel devices. IBM had had ECC (Error Correcting Code, extra bits to allow you to detect and repair bit errors) built into their mainframe processors and its memory for years. this is nothing new. Early PC’s also had parity memory bits that allowed them to detect (but not correct errors. They were eventually removed because folks got more upset with the machine stopping than having an error or two in the data, it was also an extra piece of hardware to break)

Comments are closed.